services are likely to involve the specialized treatment system as well as the primary health care system. Different assessment tools are available, including the Addiction Severity Index (McLellan et al., 1992) and the Patient Placement Criteria developed by the American Society on Addiction Medicine (CSAT, 1994).
National data sets and data sources available at the community level can be used to assess the prevalence of substance abuse problems and the services needed within health plans. Information on the chronic medical conditions resulting from alcohol and drug use or abuse is mainly available from hospital discharge surveys and death certificates (NIAAA, 1990). The major strength of these measures is their availability at the county level. The causes of death that the National Institute on Alcohol Abuse and Alcoholism (NIAAA) argues are actual measures of chronicity are cirrhosis (including chronic liver disease and cirrhosis and portal hypertension), alcohol dependence syndrome, and alcohol-related psychoses.
Data are also available on the deaths due to alcohol-related incidents, such as drunk driving. In using such alcohol-related mortality measures, NIAAA notes that factors in addition to mortality rates should be considered (NIAAA, 1991). Such indicators include the size of the population; the existing treatment capacity, including the geographic dispersal of that capacity; the amount of financial support per treatment modality; the level of urbanization; the sociodemographic characteristics of the population, such as ethnicity and age; and the existence of waiting lists for treatment programs. NIAAA claims that the data can be used to project population estimates of need by linking data on current resources with these types of data listed above in multivariate models (NIAAA, 1991 ). The major usefulness of adapting large data systems to measure chronicity is their potential usefulness in providing ratios of the number of cases to the overall prevalence, adapted for differences in population characteristics.
It is important to note the differences between the results drawn from data collected from the general population and those collected from populations in treatment systems (Corty and Ball, 1986; Rounsaville and Kleber, 1985; Weisner et al., 1995b). Although both data sources are crucial to estimating need and to developing systems for monitoring the care of chronic substance abusers, they cannot answer the same questions. Systems that track the prevalence of substance abuse in the general population provide both prevalence estimates and the need for services in the population as a whole. Data from treatment agencies provide information on trends in the group receiving the services and the needs of the individuals in that group.
The preferred data for measuring prevalence and monitoring the effectiveness of managed care organizations in responding to substance abuse problems involves epidemiologic surveys. However, these are expensive, and it is not fea-